guide
Checking a CFD loss-warning percentage without turning it into a ranking
Record the entity, date and population behind a provider's loss warning, then separate what the percentage measures from what it cannot establish.
By Betsafe Reviews Editorial Team ·
Updated Oct 5, 2026
- Document date
- Entity and product
- Limits
A loss-warning percentage is evidence about a defined set of accounts, not a broker safety score. Betsafe's checking workflow records the warning with its company, product context and observation date. Copying a number without those fields makes it difficult to tell whether two statements conflict or simply describe different scopes. This guide explains the documentary check; it does not independently audit any provider's underlying account data.
Start with the calculation's population
The FCA's relevant UK retail rules require the calculation every three months over the preceding twelve months. It includes realised and unrealised net results and charges, excludes deposits and withdrawals, and excludes accounts without an applicable open position in that period. Those boundaries matter: this is not the percentage of losing trades, website visitors or all people who have ever used the brand.
Make a dated capture that another reader can reproduce
| Capture | Reason |
|---|---|
| Exact page and company | A group can operate more than one regulated entity |
| Product and customer scope | Retail CFD and other business lines are not interchangeable |
| Displayed figure and observation date | A changing figure needs a time label |
| Calculation period if disclosed | Viewing date alone does not identify the underlying interval |
| Matching legal and risk pages | Differences need explanation before comparison |
For a concrete documentary example, CMC's UK margin page displayed a 69% loss warning when checked on 5 October 2026. That is a dated provider display, not our recomputation, an immutable figure or an endorsement. A future capture may differ. Retain the old observation rather than silently rewriting it to make an earlier research report look current.
Check the inference with an invented example
Imagine a defined sample of 100 accounts in which 70 finish below zero. A count-based statistic is 70%. It does not reveal whether each lost one unit or thousands, whether the remaining accounts made a profit or broke even, or how any one account will perform next year. If another sample has 60 losing accounts, that alone cannot identify the safer provider: customer behaviour, products, periods and sample composition could differ.
Likewise, subtracting the warning from 100 does not automatically produce a success rate. The complement is not a promise of profit. Nor should the percentage be entered into an editorial score by an improvised formula. Any scored criterion would need an explicit site methodology, appropriate evidence and approval. Our unscored documentary company checks do not acquire a rating from this number.
Handle disagreements as evidence questions
If two pages disagree, first establish that they refer to the same entity and product. Compare dates and language versions, then retain both captures and identify the missing explanation. A stale page, a changed calculation or a regional redirect is a possible cause; none should be asserted without evidence. Ask for clarification through a verified company contact if the number matters to a decision, rather than inventing a single definitive percentage.
A properly labelled warning communicates substantial leveraged-trading risk. It does not verify execution quality, withdrawals or compensation eligibility. This software-assisted guide explains how to read the evidence; no customer outcomes have been independently tested. Keep the contract's date and scope
Article sources
| Document | Publisher |
|---|---|
| COBS 22.5: retail CFD requirements ↗ | Financial Conduct Authority |
| UK margin page and displayed risk warning ↗ | CMC Markets |
| UK retail CFD policy statement ↗ | Financial Conduct Authority |